Cognitive intelligence platform for distributed M2M/ IoT systems

Abstract

Systems and methods to leverage and manage data and knowledge in a M2M or Internet of Things (IoT) network are provided. More particularly, a cognitive intelligence platform for an IoT network that provides autonomic decision support system at or near real-time and executes a dynamic runtime is provided. The hardware, software and communication design of the platform replicates the structural and operational model of the human nervous system to achieve cognitive intelligence through adaptation, collaborative learning, knowledge sharing and self-adjustment. Further, in one embodiment, the cognitive intelligence platform has three logical processing layers of increasing complexity, each of which has agents that use statistical and machine learning techniques and algorithms to resolve situational needs and update knowledge. Furthermore, each processing layer of the platform has a basic level of intelligence and additionally the hierarchy of layers aggregates the learning and intelligence at each layer.

Description

FIELD OF THE INVENTION[0001]The present invention relates generally to intelligent wireless and wireless sensor network (WSN) communication design and distributed mobile, machine-to-machine (M2M), and internet of things (IoT) system architecture.BACKGROUND[0002]With the proliferation of wireless infrastructure, smart controllers, and smart devices, numerous industries are relying on distributed machine-to-machine (M2M) environments and internet of things (IoT) for improving operational efficiency. Existing M2M deployments require numerous controllers and heterogeneous hardware / communication interfaces to interface the M2M sensors to their respective central units to receive, process and transmit data. In any physical environment, the deployed M2M / IoT platform collects structured and unstructured data and a variety of related and unrelated perpetual events which are processed by one or more nodes depending on its physical and functional capabilities.[0003]The controllers and interfac...

AI Technical Summary

Benefits of technology

[0007]According to one exemplary aspect, the disclosure relates to a cognitive intelligence platform for distributed M2M and IoT implementations that provides autonomic and real-time or near real-time decision support system. The cognitive intelligence platform has a unique hardware, software, and communication design that replicates the structural and operational model of a centralized nervous system or neural network of vertebrates, such as the human nervous system, through mutual adaptive collaborative learning, adaptive knowledge sharing, persisting and applying incremental knowledge contextually, and self-adjustment through mutual sharing of responsibilities. It is based on a dynamic run-time that has the ability to resolve the context of a changing physical and operational environment through collaborative operations which continue to mature over time. The system is able to self-adjust the entire ecosystem to adapt to changing conditions and serve real-time operational efficiency.

[0008]According to another exemplary aspect, the platform design designates operational capabilities, dependencies and hierarchies to process events, execute operations and deliver supports. The cognitive intelligence platform has three logical processing layers—edge controllers (EC), terminal aggregators (TA), and an intelligent service management hub (ISM or Hub). The edge controllers act as receptors for specific operations. They retrieve the raw events from the physical environment and broadcast them to peers and terminal aggregators to support context specific actions. The terminal aggregators supervise and support the operation control and knowledge distribution for edge controllers, and facilitate collaboration in or near real-time across the layers horizontally and vertically. They are connected to other TAs and ECs for information exchange. They are also connected to the ISM for exchange of commands and operational data. The Intelligent Service Management Hub monitors operations and uses complex statistical and machine learning techniques and algorithms to resolve situational needs and update rules for subsystems on the fly. The Hub may be deployed in one or more servers.

[0009]According to one aspect of the disclosure, the dynamic run-time behavior of the disclosed platform is accomplished in two steps—each node has a basic level of intelligence and rules to process content, data, events, and collaborate with other nodes. Additionally, the hierarchy of layers allows the platform to aggregate the learning at each level and adjust the responsibilities. The disclosed cognitive platform decouples the rules from the processors, the tasks from the workflow subroutines / agents, and delegates the processing of the intelligence into the logical layers to have autonomy in decision support and adapt changes automatically. The agents can reorganize themselves within the runtime environment to improve operational autonomy across the layers. The agents may move across the processing layers as needed depending on the capabilities of the agent and the requirements of the system. The movement or deployment of the agents is facilitated by the runtime to improve the real-time execution. The movement or deployment of the agents may be dynamic or manual. All the layers have uniform hardware and software architecture so that the system can distribute the agents, update rule engines, and translate the changing runtime semantics into a permanent operational environment. All nodes, including controllers, aggregators and ISM servers, are preloaded with minimum software (agents, rules engine, and event bus frameworks) and the ISM hosts a repository of agents. Within the connected environment, each agent sends real-time notifications to its peers and supervisor layers to seek support to handle exception conditions. The response from the supporting nodes is an immediate action and / or change in the processing rules or learning patterns to support the conditions as well as deploying any additional agents to act on those situations. The platform architecture supports the system to define the priority and order of such changes.

[0010]According to another aspect of the disclosure, the disclosed cognitive platform has a collaborative runtime environment where each node is a participant. This supports mutual learning and sharing and results in new knowledge for the system. This allows the system to assess, resolve and process the context and content dynamically and deliver adjusted actions and services in isolation and / or in collaboration with other nodes and provide real-time operational intelligence. Thus the distributed platform grows the knowledge of the platform and helps the system gain autonomy by enhancing situational awareness and decentralized decision-making abilities over time.

[0011]According to another aspect of the disclosure, the delegation of responsibility and dynamic organization allows the M2M / IoT implementations to scale horizontally and vertically without effecting existing implementation. The disclosed cognitive platform has multiple applications as embodiments which are not limited to any particular type of work performed by the work environment. For example, it may be used in smart grid asset management, connected health and remote patient monitoring, intelligent plant management, predicting maintenance operations, manufacturing control, intelligent asset and people tracking operations, real-time multi-missile combat systems, and automated high frequency trading. It may also be used, in a connected health environment, to remotely track a patient's heart rhythm and / or glucose levels and provide appropriate alerts and actions when combination of vitals shows any anomaly or is developing an unknown outcome. In such an instance, the proposed system may acquire intelligence from readings of other patients going through similar therapies, monitored through a central hospital diagnostic center, and use that acquired intelligence to dynamically improve the patient's outcome by offering predictive and preventive suggestions and notifications to patients and doctors, reduce the number of doctor visits, help on-demand diagnosis, and perform other hospital functions. As another example, the proposed system may be used in smart-grid monitoring for self-filling of components / actions. Within an unpredictable portfolio of grid connected renewable resources, an inverter monitoring controller may notify and collaborate with substation control environments, switch to distribute loads optimally to avoid overloading, or minimize loss—a situational awareness to optimize demand and distribution of energy in real-time or near real-time. As yet another example, the proposed system may be used to achieve improved efficiency and cost savings in a distributed environment of thousands of trailers each utilizing a broadband transmission by instead employing a terminal aggregator at each trailer and thus only a single broadband for the whole environment and resolve any diagnostic issues with a trailer through local collaboration. The proposed cognitive platform stores pre-fed information and also grows the knowledge in real-time by adding new knowledge or updating the existing knowledge. It processes data and shares knowledge in real-time or near real-time. The distribution of the new knowledge or updated knowledge may be in real-time or in batch mode depending on the system's need, availability of information, nature of communication, and priority of execution. While each node has a threshold of knowledge that it can store in its database, it has access to a growing knowledge base from its peers and supervisors. The proposed system intelligently addresses mission critical operations locally without depending on a centrally located computing infrastructure and thereby minimizes communication and response latency. Further objects and advantages of the invention will become apparent from a consideration of the drawings and the ensuing descriptions.

Problems solved by technology

Due to the constraints of processing and communication capacity, any single node with preloaded structural intelligence will have limited scope to correlate and aggregate raw events, develop situational awareness and take prompt actions to manage the disruptive operational environment of a M2M / IoT ecosystem.

These systems have limited ability to leverage machine intelligence and collaborative learning within a specified system boundary.

The cognition process in these systems is generally confined to data mining and based on limited pre-fed intelligence.

Images